Don Labbe, IOM Invensys Operations Management, USA
Coal supplies have widely different sulphur and nitrogen content. In the drive to lower nitrogen oxides (NOx) and sulphur dioxide (SOx) emissions in the US, coal utilization has shifted to predominantly low-sulphur and low-nitrogen Powder River Basin (PRB) coal. These coals have proven extremely effective in meeting state and federal emission requirements, but typically require furnace, fan and mill hardware modifications to maintain load generation rates.
PRB coals have the disadvantages of a higher water content, a lower heating value and in some cases longer transportation routes than local bituminous coal. Coals that have minimal water content, such as bituminous coal, also provide a carbon advantage, that is lower carbon diOxide (CO2) production per unit of generation. In a simple analysis, the potential reduction in CO2 approaches 8.5 per cent by replacing PBR coal with local bituminous coal considering coal transport.
Mixing coal sources to achieve lower CO2 however, challenges unit operations from coal handling to coal mills, furnace fuel and air controls, boiler steam temperature controls, soot blowing, precipitators, scrubbers and ash handling. To successfully manage blending coal supplies unit controls must be flexible and adjust to the requirements of the mixed fuel. However, the relative proportion of PRB that can be replaced is dependent on several factors including the performance capability of NOx/SOx emissions control systems and boiler/furnace operational flexibility.
In a study we conducted, which measured the operation of a unit burning a blend of PRB coal and bituminous coal, ranging from 25-75 per cent PRB, furnace optimization was shown to provide significant NOx reductions with coal blends. Combining SCR design margin with furnace optimization may allow coal blends to attain emissions objectives, thereby enhancing feasibility of blending.
Since burning PRB coal requires more coal and air throughput than bituminous coal for the same generated load, the mill power requirement and ID fan demand are typically reduced as the PRB coal fraction is lowered. This suggests that coal mills and fans should not be limiting factors in transitioning from PRB to a PRB/bituminous mix.
The next challenge in transitioning to a higher proportion of bituminous coal is achieving SOx emissions requirements. If the scrubber system has the capacity to handle higher sulphur coals, then attaining SOx emissions with a coal blend should be readily feasible. If the SOx scrubber system was designed for low sulphur PRB coal, then the proportion of bituminous coal may be very limited at high loads. In either case optimization of the SOx scrubber system could increase the proportion of higher sulphur bituminous coal.
Both wet scrubbers and dry scrubbers have key control variables that can improve the SOx capture rate providing an optimization opportunity for improved SOx reduction, for example slurry concentration and approach temperature for dry scrubbers. Bag house systems, while primarily designed for opacity control, also provide significant SOx capture capability when combined with dry scrubbers or chemical injection. Optimization of bag house control to increase the thickness of the ash cake and/or chemical injection may further enhance SOx capture rate and expand the allowable range of sulphur in the feed coal.
To further expand the CO2 reductions, the use of coal blends favouring bituminous coal at low and intermediate loads may be possible. Since the SCR and FGD systems are sized for full load these systems should have high operating margins at lower loads. However, adjusting coal blends quickly to meet load dispatch objectives may require increased flexibility in the coal source supplying each mill. One approach is to dedicate each mill to a specific type of coal.
At low loads the mills supplying bituminous coal can supply the majority of the coal. As a NOx or SOx emission limit is approached with increasing generated load, the mills supplying PRB coal can assume a higher proportion of the total coal supply. The appropriate proportions can be adjusted through the optimization system.
Furnace performance optimization
The in-furnace combustion characteristics of PRB coal are different from bituminous coal. These differences may shift furnace fouling rates and location of deposits, shorten flame ignition points and promote flame impingement on walls, alter superheat and reheat energy distribution, promote boiler tube failure rates and other adverse factors. The further need to lower furnace NOx emissions through furnace optimization places great importance on maintaining satisfactory furnace conditions over the spectrum of coal utilization.
The key to reducing furnace NOx emissions is controlling air/fuel ratios in the combustion zone. A typical coal fired furnace control applies air flow distribution through air dampers to influence these factors in addition to selecting coal mills and coal feed rates.
The air/fuel ratios also have a significant influence on furnace fouling, flame ignition points and impingement, burner zone reducing environment and heat flux related to tube failures. There is also an impact on energy distribution and steam temperatures.
One method to optimize furnace operation is to apply sensors in the firing zone that measure key parameters and provide feedback to the optimization system. Combining visual furnace inspections during initial commissioning of these sensors define the reasonable bounds or constraints for these sensory inputs. By incorporating these sensory inputs and their constraints into the optimization system, the furnace can achieve the desired objectives of NOx reduction coupled with maintaining satisfactory furnace conditions.
An alternative approach is to apply virtual burner zone sensors based on the primary measurements and controls of fuel and air distribution. Such virtual sensors have been applied with great success over many years in furnace optimization. The virtual sensors dynamically track load, coal feed changes and fuel quality changes providing the feedback to modulate air dampers and coal feeder bias to maintain satisfactory furnace conditions while minimizing NOx formation1.
If a coal blend is used then the burner zone virtual sensors would require information on the coal blend in each coal mill. If each mill supply silo has a specified coal type, then the operator can specify the coal through the DCS interface and thereby download the coals characteristics to the burner zone virtual sensors. If the coal supply silo is undergoing a change from one type of coal to another, the operator can specify the loading coal type.
Applying residence time models in conjunction with mill measurements, the transition in coal type should be accurately determined by the coal property portion of the burner zone virtual sensors. If pre-blended coal of unknown proportion is fed to the mill silo, then mill measurements should provide sufficient information to approximate the coal blend with an estimated uncertainty. The coal blend characteristics and related uncertainty are applied to adjust the operating constraints of the burner zone virtual sensors in a conservative direction to maintain satisfactory furnace conditions at the expense of slightly higher NOx.
Burner zone virtual sensors provide a reliable method to monitor and control furnace conditions, avoiding harsh furnace conditions. Through the application of coal type assessment for each coal mill, these virtual sensors adapt to the coal blend and provide the necessary feedback to maintain satisfactory furnace operation over the range of load.
Boiler Performance Optimization
The distribution of energy within the furnace has a major impact on the efficiency of the boiler and the turbine. For maximum overall cycle efficiency, superheat and reheat steam temperatures should approach set-point, while furnace exit gas temperatures should be as low as practical. Most boilers have limited means to adjust energy distribution.
In once-thru units there is more flexibility in the energy distribution, since the heating process is continuous from a sub-cooled water to a superheater outlet. One control challenge is the balance between superheat and reheat steam temperatures. Excessive energy distribution to the reheat portion of the furnace may result in the need for reheat sprays and thereby suffer a cycle performance penalty. Too little energy to the reheat section results in cooler reheat temperatures and a drop in turbine cycle efficiency.
For drum type units, the distribution of energy is more complex because of the fixed surface areas for evaporation, superheat, reheat and economizer functions. Maintaining the appropriate balance of energy distribution between the sections is typically a prime objective of a heat rate performance optimization system including soot blower optimization.
Figure 1: Superheat and reheat steam temperatures prior to optimization
Figure 1 illustrates the steam temperature performance of an 850 MW coal fired drum type unit prior to the installation of an optimization system2. The unit has tangentially fired drum type twin furnace boilers with eight coal mills supplying PRB coal. Design turbine throttle conditions are 1000 °F/1000 °F (538 °C/538 °C) and 2400 psig (156.5 bar). The trend displays load, the A/B side superheat steam temperatures and A/B side reheat steam temperatures. The ranges of each trend are noted below the tags. The values in white correspond to the vertical time line or final time. A large A/B split in temperatures is apparent.
Figure 2: Superheat and reheat steam temperatures following optimization
Figure 2 (on page 73) illustrates the steam temperature performance of the same unit after the installation of an optimization system. The optimization system combines furnace optimization, soot blow optimization and model predictive steam temperature control to achieve greater side to side (A/B) balance for higher average steam temperatures while lowering peak steam temperatures.
Figure 3: NOx and heat rate benefits following optimization
The higher steam temperatures contributed to significant heat rate improvements as illustrated in Figure 3. Lowering the peak steam temperatures reduced the high temperature stress and creep for the superheat and reheat sections resulting in a lower tube failure rate and higher unit availability. Most significantly, the 0.67 per cent heat rate improvement corresponds to a $700 000 fuel savings, assuming $2.25/MMBtu and a 60 per cent capacity factor. The furnace optimization also substantially lowered NOx emissions apprOximately 25 per cent.
The furnace optimization system included burner zone virtual sensors to maintain satisfactory furnace conditions while lowering NOx emissions. These virtual sensors tracked load dispatching and mill starts and stops to provide continuous furnace monitoring and feedback to the optimization system. The control of steam temperatures on this unit was particularly challenging due to an undersized superheater causing superheat temperatures to operate below set-point during steady high load conditions. When operating below set-point the superheat spray flow is shut-off and steam temperatures are prone to rapid change during load ramps.
Although little control action can be taken to curtail temperature drops, the model predictive steam temperature control proved extremely effective in limiting peak steam temperatures through precise regulation of superheat sprays and burner tilts. An additional benefit was a 25 per cent increase in the unit ramp rate, since the prior steam temperature variations were limiting unit ramp rate. Similar successes in NOx reduction, heat rate improvement, steam temperature control and ramp rate were achieved at the sister units through the application of optimization systems3.
Since bituminous coal has a higher heating value than PRB and requires less air, the water wall section of drum type units burning bituminous coal has a higher heat flux. Burning bituminous coal in a unit designed for PRB shifts the balance of energy towards the evaporation portion of the boiler. This typically lowers superheat and reheat steam temperatures, unless there is significant margin in superheat and reheat tube surface area.
Soot blower optimization and stoichiometric firing through furnace optimization have proven effective in shifting energy distribution towards the needed area. If superheat and reheat temperatures are low, then shifting energy from the water wall section to the superheat and reheat passes would be an objective of these optimization systems.
If steam temperatures drop below setpoint during steady load operation, then the need for advanced control strategies becomes more pressing. Model predictive steam temperature control provides a means to precisely control peak steam temperature during dispatch operation. This approach provides tube protection for the superheat/reheat sections supporting rapid dispatch rates while maximizing unit heat rate.
Enhancing ramp rate extends beyond steam temperature and boiler controls. Issues such as rate of change of pressure for thick walled vessels and turbine life expenditure may need to be addressed. Advanced control strategies combined with furnace optimization may prove very helpful as evidenced by the ten per cent/min dispatch rate achieved by a gas fired drum unit4.
For coal fired power plants located in the central and eastern side of the US that have installed NOx and SOx emissions abatement equipment substantial CO2 reduction approaching eight per cent may be achieved by shifting from PRB coal to more local bituminous coal. The range of the shift to bituminous coal that is feasible is dependent on the design margin of the emissions abatement equipment, but can be expanded by the application of optimization technologies. Furnace optimization provides NOx reductions that can partially offset NOx increases resulting from the shift from PRB to bituminous coal, while maintaining satisfactory furnace conditions.
This approach provides a near-term opportunity for substantial carbon reduction, while simultaneously enhancing unit dispatch capability.
1. Labbe D. Roberts D. & Brown J., Equipment Upgrades and Phased Optimization Enhance Unit Performance, 18th Annual Joint ISA POWID/EPRI Controls & Instrumentation Conference, Phoenix, AZ, June 2008.
2. Labbe D. Coker S. & Speziale A., Entergy Independence NOx/Heat Rate Optimization and Steam Temperature Control with Neural Net/Model Predictive Control Combo, 15th Annual Joint ISA POWID/EPRI Controls & Instrumentation Conference, Nashville, TN, June 2005.
3. Labbe D. Hocking W., Ray W., Anderson J. & Klepper P., Dynamic NOx/Heat Rate Optimization – Update, ISA EXPO Conference, Houston, TX, October 2006.
4. Labbe D., Runkle D., Lax J. & Chapa R., Optimizing Turbine Life Cycle Usage and Maximizing Ramp Rate, 16th Annual Joint ISA POWID/EPRI Controls & Instrumentation Conference, San Jose, CA, June 2006.
This article is based on a presentation at POWER-GEN Asia 2009, which takes place 7-9 October in Bangkok, Thailand and an ISA POWID 2009 paper (19th Annual Joint ISA POWID/EPRI Controls and Instrumentation Conference, Chicago, Illinios, May 2009).